Pre-existing diabetes is increasingly common in pregnancy and particularly prominent in low-income minority women. It is associated with an increased risk of adverse outcomes for both infant and mother. Several studies have examined the association of congenital anomalies with oral anti-diabetic drug exposure; however none have looked at the independent risk of oral anti-diabetic agents on maternal morbidities. The goal of this research is to explore the relative safety of three most commonly utilized oral anti-diabetic drug classes in pregnant diabetic women (biguanide (metformin), sulfonylurea, or thiazolidinedione). With the goal to improve maternal and child health I propose the following aims. (1) To examine annual trends in the prevalence of pre-existing diabetes in pregnant women, explore the demographic characteristics associated with pregnancy and this chronic disease state. (2) To ascertain anti-diabetic drug utilizations patterns throughout the various stages of pregnancy. (3) To compare the safety of three drug classes of oral anti-diabetic agents (ADA) exposure in the first trimester of pregnancy in terms of risk of cesarean section deliveries, other maternal morbidities and preterm birth. Specific comparisons will include use of metformin versus sulfonylureas as single ADA, or metformin versus sulfonylureas in combination with other oral ADAs. Data sources will include the Medicaid Analytic eXtract (MAX) database and the National Vital Birth Certificate Records (BCR) for Florida, Texas, New York, and Illinois from 1999 to 2006. This analysis will be restricted to women with Medicaid fee-for-service benefits linked to the birth certificate data. This linkage to BRCs enriches the MAX billing records with clinical parameters such as gestational age and the comprehensive medical history obtained through parent self-report. I will require 18 months continuous Medicaid eligibility, 15 months pre-delivery and 3 months post, and exclude women on insulin during the baseline period, as well as those who deliver multiplets from the inclusion into the pregnancy cohort. Proposed methods to address the two study aims include: calculation of annual prevalence rates of both diabetes in pregnancy and respective anti-diabetic agent utilization, linear regression to evaluate secular trends, and multivariate logistic regression analysis to determine the comparative safety of oral anti-diabetic agents in relation to cesarean section delivery, maternal morbidities and preterm birth. Common demographic variables (e.g., maternal age, education level, race/ethnicity, and geographic location) as well as clinical characteristics (e.g., hypertension, obesity, and compliance with pre-natal visits) will be used to control for confounding. The evaluation of the safety of oral anti-diabetic agents will provide better-quality perinatal care, as well as contribute to future guidance on the treatment of pre-existing diabetes in pregnancy.